Shrinkage compensation calculator

A part comes off the bed smaller than the model, because the plastic went down hot and every layer pulled in as it cooled. For most PLA prints the difference lives inside the tolerance you were going to accept anyway; for a long ABS part, or for anything that has to fit something else, it is the difference between a press fit and a paperweight. The fix is to model the part bigger by exactly the amount it will lose — and the amount is not the shrinkage percentage, it is the reciprocal of what is left. This page does that inversion and shows you the scale factor to type into the slicer. What it will not do is tell you what your shrinkage is, because nobody publishes a figure for it that would be true of your printer.

How it is calculated

modelled = final ÷ (1 − s) · scale = 100 ÷ (1 − s)

If a part keeps the fraction 1 − s of the size it was printed at, then printing it at 1/(1 − s) of the target brings it back to the target exactly. The two are not the same number and the difference grows as the shrinkage does: at 0.5% the compensation is 0.5025%, near enough to ignore, but at 20% you have to add 25%, not 20%. Adding back the same percentage you lost always leaves the part slightly small, which is why parts scaled by eye keep coming out just under.

Source: A part that keeps 1 − s of the size it was printed at has to be printed at 1/(1 − s) of the size it must end up

Questions people ask

What shrinkage figure should I put in?

One you measured. Print a rectangular test block as large as your bed sensibly allows, let it cool to room temperature, measure it with callipers along the axis you care about, and put in (modelled − measured) ÷ modelled × 100. That number is yours: it depends on the material, the nozzle and bed temperatures, whether the printer is enclosed, how large the part is and how it is oriented. This page deliberately ships no per-material defaults, because a default here would be a guess wearing a decimal point.

Do the filament datasheets not publish shrinkage?

Almost none of them do, and the ones that do are answering a different question. Of the sheets behind these pages, only Plexiwire quotes a figure — "Linear mould shrinkage 0.3-0.5 %" for its nylon PA6, measured to DIN 16742. That standard is about injection moulding: molten polymer forced into a closed steel cavity and held under pressure while it solidifies. A printed part is built in the open air, one thin layer at a time, and each layer is restrained by the one below and by the bed. The moulding figure is a lower bound and a hint about the polymer, not a number to type in here.

Why does my part shrink more in one direction?

Because it is not one material in three directions. Along X and Y the part is continuous extruded plastic and contracts as the polymer does; in Z it is a stack of layers whose height is set mechanically by the nozzle, so it barely moves at all. Long spans shrink more than short ones in absolute terms, and a part held down hard by bed adhesion is stretched while it cools and then relaxes when you pop it off. Measure the axis and the length that actually matters for the fit.

Should I scale the whole model or edit the dimension?

Scaling is easier and usually wrong for anything with a hole in it. A uniform scale grows the bolt holes, the wall thickness and the thread pitch along with the outside, which is fine for a display piece and not fine for a bracket that has to line up with existing hardware. Where one dimension has to be right, change that dimension in the model and leave the rest alone — this page gives you the millimetres for exactly that, alongside the percentage for when scaling really is what you want.

Can I compensate for warping this way?

No. Shrinkage is uniform contraction and a scale factor cancels it. Warping is what happens when the contraction is uneven — the bottom layers are pinned to the bed while the ones above pull in, so the corners lift and the part is no longer the shape you scaled. No multiplier fixes that. Warping is a temperature problem: an enclosure, a hotter bed, brims and slower cooling, not arithmetic.

Related tools